Diversified intrusion detection using Various Detection methodologies with sensor fusion

Various Intrusion Detection System (IDS) in the literature have shown that multiple classifier may be well versed in detecting the specific attack, but detecting different types of attack is low. In order to ensure high security this work focuses on multiple classifier fusion technique to increase detection rate. The primary role of classifier is to classify the correct and incorrect instance therefore multiple classifier design that is practical, and detects more attack by means of combining them is preferred here. To our best knowledge, this is the first design that considers multiple classifier in which all classifiers are different that detects both anomalies based and misuse based attacks. The dataset collected in a networking environment with the relatively high data density may contain attacks that assaults the system and thus violates system security. In this paper the operation of combining multiple classifiers that detects all categories of attack, from that improving the detection rate and true positive rate thereby reducing the false positive rate can be done. Decision based on threshold value and combining the classifiers result based on majority voting rule helps to increase the overall efficiency and accuracy in detecting the various categories of attack.